Ventilatory response to dynamic exercise is quantitatively evaluated to distinguish and determine significance of the fast (neural) compared with the slower (humoral) components. Random variation in retarding torque applied to a bicycle ergometer provides a nonanticipatory exercise regime. A normal human volunteer is instructed to maintain a constant pedalling rate regardless of retarding load. The random signal, generated from a band-limited white noise source, provides the nonanticipatory, nonabrupt exercise regime. Random, nonabrupt exercise is less likely to induce subjective contributions to ventilation than deterministic exercise regimes. Deterministic (e.g., sinusoidal and ramp) exercise is potentially learned by the subject with concomitant anticipatory breathing patterns whereas abrupt transitions (step and pulse) potentially startle, or at least momentarily alter pedalling frequency. Comparison of hypoxic augmentation and hyperoxic diminution of the ventilatory response in dynamic exercise to normoxic conditions determines the significance of oxygen interaction with exercise at peripheral chemo-receptors. Dynamic (CO2 sensitivity density, to provide a basis for quantitative evaluation. An "arterialized" CO2 controller is implemented to minimize the breath-to-breath variations to test the CO2 oscillation hypothesis. Physiological description of ventilatory control during exercise in the normal is necessary prior to assessment of pathological alteration.